摘要
针对多源异构数据并行聚类算法存在并行聚类处理精度低、处理时间长的问题,提出基于频繁项集的多源异构数据并行聚类算法;通过确定数据库内频繁项集与支持度的数值,采用关联规则满足最小支持度阈值以及最小置信度阈值,利用极大元法挖掘最大频繁项集,构建相异度数据结构矩阵;利用平均加权法获取数据库内多源异构发射数据包,使用时间窗口和频繁项集挖掘出多源异构数据特征,获取信道传输功率谱密度;利用时间反转处理以及高维相空间重构方法,实现多源异构数据并行聚类。结果表明,该算法的多源异构数据并行聚类处理精度较高,能够有效缩短处理时间。
Aiming at the problems of low parallel clustering processing accuracy and long processing time of parallel clustering algorithm for multi-source heterogeneous data,a parallel clustering algorithm for multi-source heterogeneous data based on frequent itemsets was proposed.By determining values of frequent itemsets and support in the database,association rules were used to meet the minimum support threshold and the minimum confidence threshold,and the maximum frequent item sets were mined by using maximum element method to construct a dissimilarity data structure matrix.Average weighted method was used to obtain multi-source heterogeneous emission data packets in the database.A time window and frequent it emsets were used to mine characteristics of multi-sou rce heterogeneous data to obtain channel transmission power spectral density.Using time reversal processing and high-dimensional phase space reconstruction method,multi-source heterogeneous data parallel clustering was realized.The results show that the multi-source heterogeneous data paral lel clustering processing accuracy of the proposed algorithm is high,and the processing time can be effectively shortened.
作者
赵春霞
赵营颖
宋学坤
ZHAO Chunxia;ZHAO Yingying;SONG Xuekun(School of Information Technology,Henan University of Chinese Medicine,Zhengzhou 450046,Henan,China)
出处
《济南大学学报(自然科学版)》
CAS
北大核心
2022年第4期440-443,451,共5页
Journal of University of Jinan(Science and Technology)
基金
国家自然科学基金项目(61702164)。
关键词
频繁项集
多源异构数据
并行聚类
关联规则
相异度矩阵
frequent itemset
multi-source heterogeneous data
parallel clustering
association rule
dissimilarity matrix